Extractive query-focused multi-document summarization

1. A processor-implemented method for generating a multi-document summary, the method comprising: receiving, by a processor, a query statement, one or more documents, one or more summary constraints, and one or more goals, wherein the query statement is a user-entered series of words relating to a particular topic or asks a question towards which a user desires more information, and wherein the one or more documents are selected from a group consisting of a plurality of articles, a plurality of stories, a plurality of papers, and a plurality of repositories, and wherein the one or more documents are capable of being search and summarized, and wherein the one or more summary constraints are a plurality of factors that impact a generated summary, and wherein the one or more goals are one or more criteria to be achieved when generating a summary of the one or more documents, and wherein the one or more summary constraints are fixed and must be adhered to, and wherein the one or more goals are fluid to allow for various generated summaries to each be evaluated for quality; identifying one or more keywords within the query statement, wherein each keyword relates to an important aspect of the query statement; performing a sentence selection from the one or more documents based on the one or more identified keywords, wherein performing the sentence selection comprises parsing each received document to identify one or more sentences relevant to the query statement, and wherein the performed sentence selection is a first step of a constrained global optimization problem to identify a subset of one or more sentences in each document that maximize a given quality target function, and wherein the performed sentence selection utilizes a sentence filtering technique, and wherein the sentence filtering technique is selected from a group consisting of keyword identification and similarity score rating; generating a plurality of candidate summaries of the one or more documents based on the performed sentence selection, the one or more goals, and a fully-polynomial randomized approximation scheme (FPRAS) cross entropy method; calculating a quality score for each of the plurality of generated candidate summaries using a plurality of quality features, wherein the calculated quality score is an integer between zero and one hundred or a letter grade; and selecting a candidate summary from the plurality of generated candidate summaries with the highest calculated quality score that also satisfies a quality score threshold.